Abstract Significance: The goal of this proposal is to provide drug researchers with compelling new tools that address off-target selectivity in the rational design of drugs. To do this, we propose to employ our large repository of maps of where chemical fragments tightly bind to 1,000?s of proteins (Boltzmann maps) in searches for differ- ences in spatial binding patterns of fragments across protein families. This will empower medicinal chemists in attacking the off-target toxicity challenge that plagues clinical candidates in drug discovery. Our team is currently supported by an NIH Phase II SBIR grant (2R44GM109549) to computationally produce BMaps on a large scale (1,000?s of fragment maps on 1,000?s of proteins) for fragment-based drug design (see www.boltzmannmaps.com). Building upon this, the need is to build fast tools for the design of selective com- pounds that search these fragment maps in comparing binding patterns across a large number of proteins. Putting this capability in the hands of medicinal chemists across the industry via the Web has the potential to significantly improve those clinical outcomes negatively impacted by off-target toxicities and speed the delivery of new medicines to patients. Innovation: Comparing chemical fragment binding across a large number of diverse isozymes within protein families, based on our unique repository of fragment binding maps, is a new scientific approach to the rational design of selectivity. To efficiently query our large repository for binding patterns requires that we devise a new geometric search algorithm. Aim 1: Develop a novel algorithm for comparing fragment binding patterns using geometric hashing, supple- mented with other parameters such as the relative free energy from the chemical potential ranking and other physiochemical properties of the local binding site. We will provide a Web-based interface for requesting and visualizing search results. Aim 2: Validate the tools developed under Aim 1 in a proof-of-concept project to assess selectivity of a proprie- tary set of 127 compounds across the sirtuin family of proteins confirming experimental data. Overall Impact: In summary, having a body of diverse fragment binding data across isozymes within protein presents a unique opportunity to attack the selectivity problem in drug discovery. By comparing fragment in- teraction patterns across a large number of proteins, never feasible before, an assessment of differential bind- ing between proteins to modulate and proteins to leave unaffected is now more practical. As widely adopted, the service would provide a key tool to reduce toxicities due to off-target interactions, resulting in improved success rates of clinical trials.